Septins mediate a microtubule–actin crosstalk that enables actin growth on microtubules

Significance

Cellular morphogenesis and processes such as cell division and cell migration require the coordination of the actin and microtubule cytoskeletons. Despite its broader physiological significance, actin–microtubule crosstalk is molecularly and mechanistically poorly understood. Actin–microtubule crosstalk has been viewed primarily as the regulation or guidance of microtubule dynamics by actin filaments. Here, we have discovered a mechanism by which actin filament growth is guided by microtubules. We report that septins, a poorly understood component of the cytoskeleton, can capture and link the polymerizing ends of actin filaments to microtubule polymers. We present evidence that this septin-mediated mechanism is critical for the morphology of growth cones, which are structures that direct the growth of neuronal axons through actin–microtubule crosstalk.

Abstract

Cellular morphogenesis and processes such as cell division and migration require the coordination of the microtubule and actin cytoskeletons. Microtubule–actin crosstalk is poorly understood and largely regarded as the capture and regulation of microtubules by actin. Septins are filamentous guanosine-5'-triphosphate (GTP) binding proteins, which comprise the fourth component of the cytoskeleton along microtubules, actin, and intermediate filaments. Here, we report that septins mediate microtubule–actin crosstalk by coupling actin polymerization to microtubule lattices. Superresolution and platinum replica electron microscopy (PREM) show that septins localize to overlapping microtubules and actin filaments in the growth cones of neurons and non-neuronal cells. We demonstrate that recombinant septin complexes directly crosslink microtubules and actin filaments into hybrid bundles. In vitro reconstitution assays reveal that microtubule-bound septins capture and align stable actin filaments with microtubules. Strikingly, septins enable the capture and polymerization of growing actin filaments on microtubule lattices. In neuronal growth cones, septins are required for the maintenance of the peripheral actin network that fans out from microtubules. These findings show that septins directly mediate microtubule interactions with actin filaments, and reveal a mechanism of microtubule-templated actin growth with broader significance for the self-organization of the cytoskeleton and cellular morphogenesis.

Cell morphology, dynamics, and processes such as mitosis and migration require crosstalk between the actin and microtubule cytoskeletons (1–3). Actin–microtubule crosstalk encompasses the physical crosslinking of actin and microtubule polymers and the spatial coordination of their organization and dynamics (1, 2). Regulation and guidance of polymerizing microtubules by actin filaments have been established as a principal mode of actin–microtubule crosstalk (3). Actin-binding proteins, which interact directly or indirectly with microtubule plus ends, capture and guide microtubules at sites of cortical actin and into actin-based protrusions such as filopodia (4–6). In addition, actin suppresses microtubule growth at the centrosome and can sever microtubules which are linked to the actomyosin of lamellipodia (7, 8).

Regulation of actin organization and dynamics by microtubules is a mode of actin–microtubule crosstalk that is less evident and understood. Microtubules can regulate actin organization indirectly by harboring signaling factors (9, 10). Recent work has unexpectedly revealed that linear and branched actins grow directly from microtubule plus ends through formins and the adenomatous polyposis coli (APC) (11–13). Coupling of actin filaments to microtubule plus ends can also enable the transport of actin filaments (14, 15). Actin nucleation from microtubule plus ends promotes the maintenance of actin filaments at focal adhesions (13) and might be critical for the steering of the growth cones of neuronal axons, which depends on microtubules (16, 17). In spite of the recent findings that link microtubule plus ends to actin nucleation and transport, it is unknown if and how the microtubule lattice functions in actin assembly and organization.

Septins are a family of GTP-binding proteins that assemble into higher-order oligomers and polymers, which comprise the fourth component of the cytoskeleton after actin, microtubules, and intermediate filaments (18). Septins associate with subsets of actin filaments and microtubules (19). Septins have been shown to directly crosslink actin filaments and bundle microtubules (20–23), but it is unknown if septins can concomitantly interact with both actin and microtubules, linking their organization and dynamics. Here, we report that septins provide a mechanism of actin–microtubule crosstalk, mediating the capture and growth of actin filaments on microtubule lattices and show that septins are an essential component of the actin–microtubule coordination that underlies the morphodynamics of neuronal growth cones.

Results

Neuronal growth cones are an ideal model system for studies of actin–microtubule crosstalk (1, 2, 16). At the tip of axons and dendrites, growth cones have a fan-like morphology that consists of a central microtubule-containing domain, which is proximal to the axon shaft, and a peripheral actin-rich domain with lamellipodial and filopodial protrusions (16). Growth cone morphology and dynamics depend largely on the crosstalk between the peripheral actin network and microtubules that invade from the central into the peripheral domain (16). Using high- and super-resolution microscopy, we stained neurons for Sept7 as a common subunit and marker of septin complexes. In the peripheral domain of growth cones, fluorescence line scans showed that Sept7 colocalizes with microtubule segments that coalign with actin bundles (Fig. 1A). Superresolution structured illumination microscopy (SIM) showed that septin filaments adjoin actin and microtubules at the base of filopodia, and septin puncta localize on points of contact between the distal ends of microtubules and the proximal ends of filopodial actin bundles (Fig. 1 B and C). A similar coalignment of microtubules with actin filaments and septins was observed by SIM in COS-7 cells (Fig. 1D). In addition to its localization at longitudinal contacts of microtubules with actin filaments, neuronal Sept7 was enriched at orthogonal junctions of axonal microtubules with filopodial actin filaments (Fig. 1E). To resolve Sept7 localization at the ultrastructural level of the cytoskeleton, we performed immune-gold platinum replica electron microscopy (PREM) (24). In the growth cones of primary rat hippocampal neurons, gold-labeled Sept7 molecules localized at longitudinal contacts (Fig. 1F, arrows) and orthogonal or angular fork-like junctions of actin filaments with the lattice of microtubules (Fig. 1 F and G and SI Appendix, Fig. S1A, arrowheads). Sept7 was also enriched in filamentous hook-shaped structures, which projected from the sides of the microtubule lattice and connected with multiple actin filaments (SI Appendix, Fig. S1 B and C).

Fig. 1.

Septins colocalize with overlapping microtubules and actin filaments. (A) Images show the axonal growth cone of a rat hippocampal neuron (DIV6) stained for F-actin (phalloidin), α-tubulin, and Sept7. Line scan shows fluorescence overlap between a microtubule, F-actin, and Sept7 in the peripheral domain of the growth cone. (B and C) Superresolution SIM images of the axonal growth cone of differentiated B35 neurons, which were stained for F-actin (phalloidin), α-tubulin, and Sept7. Arrowheads point to overlapping F-actin, microtubules, and Sept7. (D) SIM image of a COS-7 cell. Arrowhead points to the coalignment of a microtubule (red) with Sept7 (blue) and actin filaments (green). (E) SIM image of an axonal shaft region proximal to the growth of cone of a differentiated B35 neuron. Arrowheads point to Sept7 at the orthogonal junctions of filopodial F-actin bundles with microtubules and a coaligning actin cable. (F and G) PREM images of the growth cones of rat hippocampal neurons (DIV4) stained for Sept7. Arrows point to Sept7, which is labeled with antibodies conjugated to colloidal gold beads (yellow) and localizes at the longitudinal contact points between microtubules (burgundy) and actin filaments (cyan). Arrowheads point to Sept7 at orthogonal and angular fork-like junctions of actin filaments with microtubules.

To test whether septins interact directly and concomitantly with both actin and microtubules, crosslinking them together, we performed in vitro binding assays using stable prepolymerized microtubules, phalloidin-stabilized actin filaments, and recombinant Sept2/6/7, the minimal septin complex which has been shown to bind separately actin and microtubules (23, 25). Using total internal reflection fluorescence (TIRF) microscopy, we assayed for the association of actin filaments with immobilized microtubules after coating them with Sept2/6/7. In the physiological range of intracellular septin concentrations (200 to 800 nM) (26), actin filaments bound microtubules in a septin concentration-dependent manner (Fig. 2A and SI Appendix, Fig. S2A). Actin–microtubule binding peaked at 500 nM of Sept2/6/7 and tapered off with higher concentrations (Fig. 2A and SI Appendix, Fig. S2A). This biphasic effect resembled the increase and diminution of microtubule plus end growth by nanomolar and micromolar Sept2/6/7 concentrations, respectively, which corelates with a transition of Sept2/6/7 from oligomers to higher-order polymers with increasing concentrations (25). Coating microtubules with mCherry-Sept2/6/7 showed that septins are present along the actin-bound microtubule lattice (SI Appendix, Fig. S2 B and C). Prebinding of Sept2/6/7 to microtubules was not required for actin–microtubule crosslinking. Mixing Sept2/6/7 with prepolymerized stable microtubules and actin filaments in solution resulted in elongated bundles consisting of both microtubules and actin (Fig. 2 B). Taken together with the intracellular localization of septins to actin-bound microtubules, these data demonstrate that septins mediate actin–microtubule interactions and, therefore, can facilitate actin–microtubule crosstalk.

Fig. 2.

Septins directly crosslink actin filaments and microtubules and recruit polymerizing actin to microtubules. (A) Taxol-stabilized microtubules were immobilized in a TIRF chamber and coated with increasing Sept2/6/7 concentrations prior to flowing phalloidin-stabilized actin filaments. Plot shows the mean (±SEM) of the average intensity of rhodamine-labeled actin per microtubule after 15 min of incubation with microtubules, which were precoated with 0 nM (n = 424 microtubules), 100 nM (n = 334), 200 nM (n = 406), 500 nM (n = 455), 750 nM (n = 433), 1 µM (n = 391), and 2 µM (n = 371) of Sept2/6/7. n.s., not significant (P > 0.05); ****P < 0.0001. (B) Images show the result from mixing in solution prepolymerized phalloidin-stabilized actin filaments with taxol-stabilized microtubules in the absence or presence of Sept2/6/7 (500 nM). (Scale bars, 20 μm.) (C and D) Time-lapse TIRF microscopy of G-actin polymerization in the presence of microtubules decorated with Sept2/6/7 complexes (500 nM). Frames show gradual decoration of microtubules with actin (C; Scale bar, 5 μm.), and the recruitment and binding of an actin filament to the microtubule lattice (D; Scale bar, 1 μm.) (E) Quantification shows percentage (mean ± SEM) of microtubule coverage with actin after 10 minutes of G-actin polymerization in the absence or presence of Sept2/6/7 (500 nM). Data were quantified from 44 to 46 microtubules from three independent experiments and analyzed with the Mann–Whitney U test. ****P < 0.0001. (F and G) TIRF microscopy frames show actin filaments that are partially attached to Sept2/6/7-decorated microtubules (F; Scale bar, 5 μm.) and undergo bouts of polymerization, generating new filaments that branch off from the microtubule-bound actin filament (G; Scale bar, 1 μm.) (H) TIRF frames show the elongation of a microtubule-bound actin filament along the microtubule lattice in the presence of soluble Sept2/6/7 (500 nM). (Scale bar, 5 μm.)

To examine whether septins can mediate interaction between polymerizing actin and microtubules, we used time-lapse TIRF microscopy to image actin polymerization in vitro and in the presence of microtubules with or without Sept2/6/7. In the presence of Sept2/6/7, microtubule lattices were gradually coated with actin (Fig. 2 C–E and Movie S1). After 10 min of actin polymerization, >60% of the microtubule length was covered with actin (Fig. 2E). In contrast, less than 10% of the microtubule lattice was decorated with actin in the absence of Sept2/6/7 (Fig. 2E and Movie S2). Actin associated with microtubules initially as small seed-like filaments and subsequently as longer filaments, which were captured onto the immobilized microtubules as they diffused into the TIRF field (Fig. 2 C and D and Movie S1). These capture events resulted in complete and stable alignment of actin filaments with microtubules (Fig. 2D) or partial attachment with the unbound segment of an actin filament branching off at an angle from the microtubule lattice (Fig. 2F, arrowheads). The latter resembled the orthogonal type of attachments of actin filaments with axonal microtubules at sites of septin enrichment (Fig. 1 E–G and SI Appendix, Fig. S1). De novonucleation and elongation of actin filaments on microtubule lattices were not evident in the spatiotemporal scale of the assay. However, microtubule-bound actin filaments exhibited polymerization dynamics, growing at an angle from the actin–microtubule bundle (Fig. 2G and Movie S3) or processively along the microtubule lattice (Fig. 2H and Movie S4). Overall, gradual decoration of microtubules with actin was the result of continuous capture of diffusing actin filaments from the fluid phase of the glass chamber and their polymerization on the lattice of microtubules. In the absence of Sept2/6/7, actin filaments bound occasionally to microtubules, but these attachments were transient and unstable, and followed by dissociation of actin filaments (SI Appendix, Fig. S3A and Movie S2).

Next, we reconstituted actin growth from immobilized phalloidin-stabilized actin seeds in vitro and imaged their interactions with taxol-stabilized microtubules that were coated with or without Sept2/6/7. We found that Sept2/6/7 increased the overlap between microtubules and polymerizing actin filaments (Fig. 3 A and B). Analysis of the collision events between growing actin ends and microtubules showed that in the absence of Sept2/6/7, actin ends crossed over microtubules with virtually no overlapping or zippering (Fig. 3C and Movie S5). On Sept2/6/7-coated microtubules, however, there was a drastic increase in the overlap of polymerizing actin ends with microtubules at collisions of <30° angle (Fig. 3C and Movie S6). Strikingly, polymerizing actin ends grew along the lattice of microtubules in a zippering manner and continued to grow after overtaking the distal microtubule end while remaining bound to the microtubule lattice (Fig. 3D and Movies S6). Prior to zippering, growing actin ends were captured at either the end or midpoint of the microtubule lattice, and Sept2/6/7 was present at these attachment points (SI Appendix, Fig. S3 B–D and Movie S7). Occasionally, more than one growing filaments zippered sequentially and in an anti-parallel orientation along the lattice of a Sept2/6/7-coated microtubule (SI Appendix, Fig. S3D). Quantification of the rate of actin elongation showed no change upon attachment and zippering along the Sept2/6/7-coated microtubule lattice (1 ± 0.03 vs. 0.97 ± 0.04 μm/min; n = 13), which suggests that microtubule-bound Sept2/6/7 associates with polymerizing actin filaments through weak and/or transient interactions that do not sterically hinder actin growth. Thus, septins can mediate a microtubule-templated actin growth, capturing polymerizing actin ends and facilitating their growth along microtubule lattices.

Fig. 3.

Septins enable microtubule-templated actin elongation. (A) TIRF images show actin filaments (green) and microtubules (red) and their overlap (gray) after 30 min of actin polymerization from immobilized actin seeds in the presence of taxol-stabilized microtubules, which were coated with or without Sept2/6/7 (0.5 μM). (Scale bar, 20 μm.) (B) Mean (±SEM) actin fluorescence (sum intensity; AU) per microtubule length (μm). Data were quantified from microtubules coated with 0 μM (n = 108) and 0.5 μM (n = 128) Sept2/6/7, and analyzed with the Mann–Whitney U test. ****P < 0.0001. (C) Plots show the distribution of the angles of collision between polymerizing actin ends and microtubules as percentage of total collisions, and their categorization into overlapping (red) or non-overlapping crossover (blue) events based on actin end coalignment with the microtubule lattice or lack thereof (n = 236 to 397 events). (D) Still frames of time-lapse TIRF imaging of the capture and zippering of a polymerizing actin end (arrowhead) along a stable Sept2/6/7-coated microtubule. (Scale bar, 5 μm.)

Capture and growth of polymerizing microtubules along actin have been the predominant mode of actin–microtubule crosstalk in vivo. Surprisingly, in vitro assays showed that actin can polymerize directly from microtubule plus ends (12), and recent work indicates that the microtubule plus end protein APC promotes actin polymerization at focal adhesions and neuronal growth cones (11, 13). Our results suggest that actin growth can also occur on microtubule lattices. To probe for events of actin growth on microtubule lattices in living cells, we used time-lapse TIRF microscopy to image the actin and microtubule dynamics in neuronal growth cones of differentiated B35 neurons that expressed GFP-Sept7. In the central domain of neuronal growth cones, we observed actin filaments growing from the sides of microtubules in an angled orientation (Fig. 4A, arrows), resembling the branch-like position and polymerization of microtubule-associated actin filaments that were reconstituted in vitro (Fig. 2 F and G). These growth events took place on microtubules that were fully embedded in the veil of the growth cone and originated from GFP-Sept7 puncta (Fig. 4A and Movie S8). As observed in the growth cones of fixed neurons by SIM and PREM (Fig. 1 B and F), GFP-Sept7 also localized to the lattice of microtubules that coaligned with actin filaments, which were labeled with F-tractin-iRFP670 (Fig. 4B, arrowheads). This microtubule-associated actin extended longitudinally along the axis of the microtubule lattice (Fig. 4B, arrows), while GFP-Sept7 continued to localize on segments of actin–microtubule overlap. Taken together with the branchlike events of actin growth, these findings indicate that septins are present on microtubule sites of actin dynamics and growth in neuronal growth cones.

Fig. 4.

Septins localize to microtubule sites of actin growth and required for protrusive growth cone morphology. (A and B) Still frames from time-lapse TIRF imaging of growth cones of differentiated B35 neurons, which expressed GFP-Sept7 and F-tractin-miRFP670 and were labeled with SPY555-tubulin. Dashed rectangles (Top Left) show the regions of the growth cone areas, which are depicted in higher magnification in the still frames. Arrows point to the ends of growing actin filaments, and arrowheads point to the GFP-Sept7 puncta that localize to sites of actin and microtubule overlap. (Scale bars, 2 μm.) (C) SIM images show representative fan- and club-shaped and collapsed growth cones of differentiated B35 neurons, which were stained for microtubules (α-tubulin; green) and F-actin (phalloidin; magenta) after transfection with control and Sept7 shRNAs. Bar graph shows the percentage distribution of growth cones (n = 21 to 24) based on their morphology. (Scale bars, 5 μm.) (D and E) Quantification of the morphology (D) and surface area (E; mean ± SEM) of neurite growth cones in primary rat hippocampal neurons, which were transfected on DIV1 with scrambled shRNA (n = 99 neurites), Sept7 shRNA (n = 113 neurites), and Sept7 shRNA with mCherry-Sept7 (n = 116 neurites). Data were analyzed pairwise with the Mann–Whitney U test; n.s.: not significant; **P < 0.01; ***P < 0.001.

Given that the morphodynamics and protrusive activity of growth cones depend on actin–microtubule crosstalk, we reasoned that septins may impact growth cone morphology. In differentiated B35 neurons, Sept7 depletion shifted the distribution of growth cones from fan- to club-shaped and fully collapsed, which phenotypically resembles the response to repulsive cues that cause GC retraction (Fig. 4 C and D) (27). Actin–microtubule was also reduced in the growth cones of Sept7-depleted neurons; the percentage of microtubule area overlapping with actin decreased from 53 ± 5% to 35 ± 3% (P = 0.016; n = 16 to 19). In embryonic rat hippocampal neurons, Sept7 depletion resulted in a similar decrease in the fan shape and surface area of growth cones, both of which were rescued upon expression of Sept7 Fig. 4E. These data reveal a hitherto unknown function for septins in actin–microtubule coordination and the maintenance of protrusive growth cones.

Discussion

Our results demonstrate that septins directly mediate actin–microtubule interactions, enabling actin capture and growth on microtubule lattices, and promoting the protrusive morphology of growth cones. The discovery of a septin-mediated mechanism of actin capture and elongation along microtubules indicates that the microtubule lattice can template actin in a similar fashion to how actin filaments capture and guide microtubule plus ends. We posit that this mechanism might be more widely utilized than hitherto known. In cell-free assays of actin and microtubule copolymerization, the microtubule-associated protein tau mediates the polymerization of single actin filaments or bundles along microtubules (28), but it is unknown if tau functions similarly in neurons. In vitro assays also show that the growth arrest-specific 2-like protein Gas2L1 promotes zippering of microtubules and actin filaments, but in neurons, Gas2L1 localizes predominately on actin filaments (29).

Septins emerge as a distinct class of actin–microtubule crosslinkers, which unlike other crosslinkers form a multimeric interface that may enable multivalent interactions with both actin filaments and microtubules. Additionally, the heteromeric nature of septin complexes allows for simultaneous attachment to actin and microtubules through different subunits (e.g., Sept2, Sept6, and Sept7), some which could bind only actin or microtubules and others may interact with both. Polymerization of actin filaments along septin-coated microtubules is indicative of transient and low-affinity interactions, which do not restrict actin growth. High-valency and low-affinity interactions are long thought to underlie the interactions of MAPs and actin (30), and the repeat motifs of MAPs such as tau appear to provide a multivalent actin–microtubule crosslinking (31). Sept2/6/7 appears to have a low affinity for microtubules and actin filaments, but the individual subunits of septin multimers can provide multivalency. Septin multimers are likely to interact with multiple actin filaments, which explains the sequential zippering of two or more polymerizing actin filaments along a single Sept2/6/7-coated microtubule. The existence of an upper septin concentration threshold, above which actin–microtubule crosslinking is abrogated, suggests that septins undergo conformational changes that hinder actin- and microtubule-binding. At higher micromolar concentrations, septins are known to assemble into higher-order paired filaments and to constrain microtubule growth, which is enhanced by septin oligomers (32). We posit that the state of multimerization of septins imparts a conformational plasticity at the interface of actin and microtubules, which can strengthen or weaken actin–microtubule crosslinking in a regulatable manner.

Our discovery of a septin-mediated mechanism of actin coupling to the microtubule lattice adds to the recent findings of actin nucleation from microtubule plus ends, and the presence of actin filaments in the lumen of microtubules (11–13, 33). We have shown that septins localize to microtubule and actin filament junctions and overlap in axonal shafts and growth cones, which morphologically and functionally depend on microtubule–actin coordination. In vitro reconstitution experiments indicate that microtubule-associated septins capture actin filaments and the ends of polymerizing actin rather than promoting de novo nucleation of actin filaments. It is plausible, however, that the actin filaments, which are linked to microtubules by septins, serve as mother filaments for the formation of Arp2/3-nucleated branches. In agreement with this possibility, we observed actin filaments growing sideways from microtubule-bound septins in vitro and in neuronal growth cones. We envisage that microtubule-associated septins may capture actin filaments, which are dislodged or severed from the plasma membrane of the axon shaft and the retrograde flow of the growth cone. This microtubule-bound actin could serve as a mother filament for nucleation/elongation factors to generate the branched actin networks that underlie the actin waves of the axon shaft, which are critical for growth cone dynamics and axon outgrowth (34–36). Capture of polymerizing actin on the surface of microtubules could also be of key importance in the trafficking and positioning of membrane organelles (e.g., endosomes and Golgi), which nucleate actin or are bound to actin densities, and their transitioning from microtubule- to actin-dependent transport (37, 38). Lastly, septins may play a key role in the coupling of actin filaments to the axonemal microtubules of primary cilia (39). In sum, the discovery of a septin-mediated mechanism of microtubule–actin crosstalk provides an insight into microtubule-based actin organization and dynamics and has broader significance for cellular and morphogenetic processes that require microtubule–actin coordination.

Materials and Methods

Primary neurons were isolated from freshly dissected embryonic (E18) rat hippocampus (BrainBits/Transnetxy Tissue), and B35 rat neuroblastoma cells were differentiated in Dulbecco’s modified eagle (DME) media containing 0.1 mM dbcAMP (Sigma-Aldrich) and N2 supplement (ThermoFisher Scientific) and plated on 1.25 μg/mL laminin and/or 1 mg/mL poly-l-lysine. COS-7 (ATCC: CRL-1651) cells were maintained in high-glucose DME (Sigma) with 10% fetal bovine serum (R & D Systems). Structured illumination and TIRF microscopy were performed on the OMX V4 DeltaVision imaging platform (GE Healthcare) with 60X/1.42 NA and 60X/1.49 NA (Olympus) objectives, respectively. PREM was performed with a JEM 1011 transmission electron microscope (JEOL) operated at 100 kV, and images captured with an ORIUS 832.10W CCD camera (Gatan). Immunofluorescence staining was performed with rabbit anti-SEPT7 (IBL America), mouse anti-tubulin (DM1a, Sigma), and iFluor488- (ATT Bioquest) or rhodamine-phalloidin (Cytoskeleton Inc). In vitro experiments were performed with rabbit skeletal muscle actin and porcine brain tubulin. Details of all the materials and methods are provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Sept2/6/7 promotes decoration of microtubules with polymerizing actin. Taxol-stabilized microtubules were immobilized to the glass surface of a flow chamber, and coated with Sept2/6/7 (500 nM). TIRF time-lapse movie shows the gradual capture and decoration of microtubules (magenta) with polymerizing actin filaments (green) after flowing an actin polymerization mix. Images were acquired at 10 s per frame. Scale bar, 5 μm. Also, see Figure 2C–D.

Movie S2.

Sept2/6/7-free microtubules are not stably coated with polymerizing actin. Taxol-stabilized microtubules were immobilized to the glass surface of a flow chamber, and imaged with TIRF microscopy after flowing an actin polymerization mix. Representative movie shows the lack of stable capture and decoration of actin filaments (green) along the lattice of microtubules (magenta). Images were acquired at 10 s per frame. Scale bar, 5 μm. Also, see Figure S3A.

Movie S3.

Actin filament polymerizes from the sides of the microtubule lattice in the presence of Sept2/6/7. Taxol-stabilized microtubules (magenta) were immobilized to the glass surface of a flow chamber, and coated with Sept2/6/7 (500 nM) prior to flowing an actin polymerization mix. TIRF microscopy movie shows a microtubule-bound actin filament that undergoes bouts of polymerization, elongating from the sides of the lattice of microtubules (magenta). Images were acquired at 10 s per frame. Scale bar, 2 μm. Also, see Figure 2G.

Movie S4.

Actin filament polymerizes along the microtubule lattice in the presence of Sept2/6/7. Taxol-stabilized microtubules (magenta) were immobilized to the glass surface of a flow chamber prior to flowing actin polymerization mix containing Sept2/6/7 (500 nM). TIRF microscopy movie shows a microtubule-bound actin filament (green) that elongates processively along the microtubule lattice. Images were acquired at 10 s per frame. Scale bar, 5 μm. Also, see Figure 2H.

Movie S5.

The elongating ends of actin filaments cross over microtubules in the absence of septins. Representative movie of TIRF imaging of actin (green) polymerizing from phalloidin-stabilized actin seeds (magenta) in the presence of taxol-stabilized microtubules (red). Images were acquired at the rate of 5 s per frame. Scale bar, 5 μm. Also see Figure 3A–C.

Movie S6.

Growing actin filament ends bind and elongate along the lattice of septin-coated microtubules. Representative movie of TIRF imaging of actin (green) polymerizing from phalloidin-stabilized actin seeds (magenta) in the presence of taxol-stabilized microtubules (red), which were pre-coated with recombinant Sept2/6/7 complexes (dark; 500 nM). Arrowheads point to microtubule sites of actin filament end docking and overtaking, marking the beginning and end of actin-microtubule zippering. Images were acquired at the rate of 5 s per frame. Scale bar, 5 μm. Also see Figure 3A–C.

Movie S7.

Actin filament growth on the lattice of a microtubule coated with mCherry-Sept2/6/7. Representative movie of TIRF imaging of actin (green) polymerizing from phalloidin-stabilized actin seeds in the presence of taxol-stabilized microtubules (red), which were pre-coated with recombinant mCherry-Sept2/6/7 complexes (cyan; 500 nM). Arrowheads point to the microtubule sites, on which actin zippering along the microtubule lattice begins and ends. Images were acquired at the rate of 5 s per frame. Scale bar, 5 μm. Also see Figure 3D.

Movie S8.

Actin filament growth at a Sept7-enriched site of growth cone microtubules. Time-lapse TIRF imaging of an elongating actin filament (cyan; F-tractin-miRFP670) that originates from a microtubule (orange; SPY555-tubulin) associated Sept7 (magenta; GFP-Sept7) in the growth cone of a differentiated B35 neuron. The movie begins at a viewing rate of 10 frames per second and slows down to 2 frames per second for visualizing the stages of actin filament growth (arrowhead). Scale bar, 2 μm. Also see Figure 4A.

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.